|Home | About | Journals | Submit | Contact Us | Français|
The occurrence of mutations in methicillin-resistant Staphylococcus aureus (MRSA) during persistent infection leads to antimicrobial resistance but may also impact host-pathogen interactions. Here, we investigate the host-pathogen consequences of 2 mutations arising in clinical MRSA during persistent infection: RpoB H481Y, which is linked to rifampicin resistance, and RelA F128Y, which is associated with an active stringent response. Allelic exchange experiments showed that both mutations cause global transcriptional changes, leading to upregulation of capsule production, with attenuated virulence in a murine bacteremia model and reduced susceptibility to both antimicrobial peptides and whole-blood killing. Disruption of capsule biosynthesis reversed these impacts on innate immune function. These data clearly link MRSA persistence and reduced virulence to the same mechanisms that alter antimicrobial susceptibility. Our study highlights the wider consequences of suboptimal antimicrobial use, where drug resistance and immune escape mechanisms coevolve, thus increasing the likelihood of treatment failure.
The widespread increase in Staphylococcus aureus antibiotic resistance has dramatically narrowed treatment choices, especially with the appearance of resistance to antimicrobials such as vancomycin and daptomycin [1–3]. Resistance often emerges in vivo during persistent infection, and a number of studies have investigated the genomic basis for this phenomenon [3–5]. Other bacterial factors have also been linked to persistent staphylococcal infections, such as reduced activity of the agr (accessory gene regulator) quorum-sensing system , persistent intracellular survival , host antimicrobial peptide (AMP) resistance [8, 9], and emergence of small-colony variants (SCVs) . Rifampicin resistance is also common in many reports of persistent S. aureus bacteremia [11–15].
Previously, we investigated the molecular basis for antimicrobial resistance and altered colony phenotype in an isogenic series of sequence type 5 (ST5) methicillin-resistant Staphylococcus aureus (MRSA) clinical isolates . The final isolate was a SCV that demonstrated attenuated virulence in an invertebrate model, profound global transcriptional changes, and enhanced capsule biosynthesis. Four single-nucleotide polymorphisms (SNPs) were detected between the SCV and the original wild-type MRSA. One SNP occurred in relA, leading to a F128Y substitution in the hydrolase domain of the enzyme. This resulted in accumulation of the alarmones guanosine pentaphosphate or tetraphosphate ([p]ppGpp) and a persistent active stringent response, a bacterial response to promote adaptation, and resilience in the face of adverse conditions [15, 16]. However, F128Y only partially explained the growth defect of the SCV isolate. We then wondered whether any of the additional SNPs, particularly an RpoB H481Y substitution, might be contributing to the full SCV phenotype. In S. aureus, rpoB mutations induce resistance to rifampicin  and have recently been shown to contribute to reduced vancomycin susceptibility . RpoB H481Y is one of the most frequent S. aureus substitutions resulting in rifampicin resistance, which occurs, in this case, through mitigation of the binding affinity of rifampicin to RNA polymerase . While such mutations lead to antimicrobial resistance, interest is emerging regarding the impact of SNPs that accumulate in clinical S. aureus isolates under combined antibiotic and host selective pressures, leading not only to antimicrobial resistance but also to altered host-pathogen interactions that favor persistent infection . For example, host AMPs are one of the first-line innate defense mechanisms against S. aureus infection, and a link between bacterial mutations associated with reduced vancomycin and daptomycin susceptibility, persistence, and reduced AMP susceptibility has been demonstrated [20, 21]. This effect might be driven by alterations in bacterial cell surface charge, but the specific mechanisms used by S. aureus to evade innate immune response, including AMP killing, are poorly understood.
Here, we made single- and double-site–directed RpoB H481Y and RelA F128Y substitutions in our parental ST5 clinical MRSA isolate and investigated the impact of these mutations on virulence, persistence, bacterial gene expression, and innate immune evasion. We show for the first time that these mutations, particularly RpoB H481Y, lead to specific transcriptional changes that alter capsule gene and agr expression, which in turn reduce virulence, increase resistance to host AMP killing, and result in persistent infection.
Bacterial strains were stored in glycerol broth at −80°C. S. aureus strains were cultured in heart infusion broth or agar (Oxoid), and Escherichia coli was cultured in LB broth or onto LB agar (Oxoid). The following antibiotics were used, if required: for E. coli, ampicillin 100 µg/mL; and for S. aureus, chloramphenicol 25 µg/mL and anhydrotetracycline 500 ng/mL. The study isolates and plasmids are listed in Table Table11.
Wild-type 6-week-old female BALB/c mice were injected via the tail vein with 100 µL of bacterial saline suspension containing approximately 1 × 108 colony-forming units (CFU) of the test strain . The mice were monitored every 8 hours until completion of the experiment, and mice showing signs of stress were euthanized by CO2 inhalation. All surviving mice were euthanized 7 days after infection. Kaplan-Meier curves were plotted and the log-rank test performed using Prism for Macintosh, version 5.0 (GraphPad Software, CA), with a significance level of P ≤ .05. Eleven mice were tested per strain. All experiments were performed in accordance with the Animal Research Ethics Committee at Monash University.
Wild-type 6-week-old female BALB/c mice were injected via the tail vein with 100 µL of bacterial saline suspension containing approximately 1 × 108 CFU of either JKD6210 (wild-type) or JKD6229 (SCV). Mice were euthanized after 18 hours (JKD6210 and JKD6229), 3 days (JKD6229), or 7 days (JKD6229), and the CFU in liver and kidney were calculated after collection into 1 mL of phosphate-buffered saline (PBS), followed by mechanical homogenization and serial dilution. Blood was collected for culture at 18 hours (JKD6210 and JKD6229) and at 2, 3, 5, and 7 days (JKD6229).
Site-directed and knockout S. aureus mutants were generated using the pKOR1 system as previously described [15, 22] and as summarized in Table Table1.1. For site-directed mutagenesis, a 2-kb fragment containing the nucleotide mutation (or complement) of interest was cloned into pKOR1. For knockout mutants, 1 kb of DNA sequence flanking the target gene was cloned into pKOR1. We generated the RpoB H481Y mutation in the wild-type parent strain JKD6210 (and in an rpoB complement of this mutant), and we made an RpoB/RelA double mutant, using our previously generated RelA F128Y mutant (JKD6301). Two capsule-deletion strains were generated in JKD6229: a capA deletion (BPH1012) and a capO deletion (BPH1046). Polymerase chain reaction and Sanger sequencing were used to confirm successful allelic exchange and deletion. All studies involving recombinant DNA were approved by the Austin Health Institutional Biosafety Committee.
Single-end genome sequencing of the mutant JKD6301 was performed using Ion Torrent sequencing as previously described .
Microarray transcriptional analysis was performed using TIGR, version 9 S. aureus arrays with total RNA extracted from exponential phase culture (optical density at 600 nm [OD600], 0.5), in triplicate, as previously described . Data were analyzed in the Bioarray Software Environment, using Bioconductor and Limma [23, 24]. P values were adjusted for multiple comparisons using false-discovery rate correction. These results for mutant strains were compared to the previously defined transcriptional profile of strain JKD6229 . A ratio of >1.5-fold with an adjusted P value of < .05 was considered statistically significant.
An Agilent Technology 1200 Series HPLC equipped with an analytical reversed-phase column (Agilent Eclipse XDB-C18 4.6 mm × 150 mm) was used for quantification of δ-hemolysin production. Culture supernatants (100 μL) were injected and eluted with a water/acetonitrile gradient (0.1% trifluoroacetic acid) from 0%–100% acetonitrile over 28 minutes with a total run time of 35 minutes, flow rate 1 mL/min and peaks quantified at a wavelength of 214 nm. The identity of the peak due to δ-hemolysin (Retention time, 26 minutes) was confirmed by electrospray ionization–mass spectrometry analysis. The relative concentration of δ-hemolysin in each isolate was determined by comparison of the peak area to that of synthetic formylated PSMa3. Because formylated PSMa3 and δ-hemolysin are similar in length (22 residues and 26 residues, respectively), the absorbance at this wavelength is comparable for both peptides. Because this work is examining the relative concentrations between strains, a peptide of similar length is an acceptable standard for concentration determination. The standard curves were constructed in the concentration range of 6.2 of 100 µg/mL and were linear over this range.
The whole-blood killing assay was modified from previously described methods . Briefly, 250 μL of washed bacterial suspension (1 × 105 CFU/mL) was mixed with 750 µL of freshly collected heparanized human blood and incubated at 37°C for 4 hours with agitation. Serial dilutions for CFU enumeration at baseline and after 4 hours were performed, and the ratio of survival at 4 hours to survival at baseline was calculated. Independent experiments were repeated 3 times with 5 different healthy donor blood samples. These studies were approved by the Austin Health Human Research Ethics Committee.
The antimicrobial activity of human β-defensin 2 (hBD-2) and human neutrophil peptide 1 (hNP-1) (Peptide Institute, Japan) was assayed by the radial diffusion method, as previously described, in pH-defined agar (pH 7.5) . At least 3 biological repeats were performed for each strain/peptide combination.
Propidium iodide (Ex535nm/Em620nm; Sigma) was used to determine membrane permeabilization of bacterial cells after treatment with antimicrobial peptides . Bacteria were harvested in the exponential growth phase (OD600, 0.5), washed 3 times with PBS, and diluted to 1 × 106 CFU/mL in 100 µL 10 mM PIPES buffer at pH 7.5. Peptide (hNP-1 or hBD-2, 20 μg/mL) was added to the suspension and incubated at 37°C for 15 minutes. The cells were stained by adding 900 µL of proprium iodide staining buffer (5.0 µg/mL proprium iodide in 50 mM K+ MEM, pH 7.2) for 10 minutes. Heat-killed bacteria (95°C for 15 minutes) and non–peptide-exposed bacteria were used as positive and negative controls, respectively. Flow cytometry was performed (LSRFortessa instrument; Becton Dickinson) in 10 mM K+ MEM, pH 7.2. Fluorescence of a minimum of 5 × 103 cells was acquired for statistical analysis. The analysis was performed in triplicate for each peptide/strain combination, and data were analyzed using FACSDiva software (BD Biosciences).
A mouse bacteremia model was used to characterize the virulence phenotype of the clinical SCV as compared to the wild-type parent strain. As shown in Figure Figure11A, the SCV isolate demonstrated a significant reduction of virulence. Clinically, this isolate was associated with a persistent infection, with >100 days of culture-positive clinical samples, despite antimicrobial therapy . The in vivo persistence of this isolate was next assessed using culture of blood and organs after murine infection (Figure (Figure11B–D), which demonstrated similar bacterial burdens in blood, liver, and kidney at 18 hours after infection, with a high bacterial burden persisting at 160 hours for the SCV.
The single mutants RpoB H481Y and RelA F128Y demonstrated reduced virulence as compared to the wild-type strain; however, despite the growth defect of the relA mutant strain, the rpoB mutant was significantly less virulent (Figure (Figure2).2). The difference in virulence between the rpoB single and the rpoB/relA double mutants was not significant. Genome sequencing of JKD6301 (relA mutant) demonstrated no additional SNPs in this strain.
Given the significant virulence reduction observed in these mutants, we next investigated the impacts of the 2 mutations on gene expression. Microarray analysis revealed significant global transcriptional changes for both the RpoB H481Y and RelA F128Y mutant, as well as for the double mutant, compared with the wild-type strain (Figure (Figure33A). A total of 183 genes were differentially regulated in the relA mutant JKD6301, and 361 genes were differentially regulated in the rpoB mutant BPH1020, compared with 890 genes differentially regulated in the SCV JKD6229 by using the same fold-ratio cutoff . Only 54 genes were differentially regulated in all 3 strains. However, the combined expression data for the single mutants showed substantial overlap with the clinical SCV transcriptional profile, suggesting a synergistic effect of the single mutations (Figure (Figure33A). Closer scrutiny of the microarray data showed significant upregulation of genes encoding capsule biosynthesis enzymes (most marked for the rpoB mutants and the SCV strain; 20–50-fold increases), significant effects on metabolism, and, intriguingly, upregulation of agr activity, which was prominent for all strains containing the RpoB mutation (Figure (Figure33B and Supplementary Table 2). Associated with the agr effects, upregulation of the 2 genes encoding phenol-soluble modulins (PSMs) was demonstrated (Figure (Figure33B). The full list of differentially regulated genes for all strains is presented in Supplementary Table 2. δ-hemolysin expression levels correlated with the array data for all strains (Supplementary Figure 1). For the SCV strain, we previously demonstrated by means of immunoblot analysis that upregulation of capsule biosynthesis genes was associated with overproduction of capsule .
To investigate the contribution of rpoB and relA mutations to in vivo persistence of S. aureus, a whole-blood killing assay was used. The clinical SCV isolate JKD6229 showed significantly higher survival than the wild-type clinical isolate JKD6210 (Figure (Figure4).4). The rpoB mutant and the double mutant demonstrated the same degree of enhanced survival, and this was reversed in the strain in which rpoB was repaired. However, the relA mutation had no impact on survival outcome. Because previous data have demonstrated a role for the S. aureus capsule in resistance to opsonophagocytic killing [28, 29], and because there was a marked upregulation of capsule biosynthesis genes (particularly for strains containing the rpoB mutation), 2 acapsular strains were generated in the SCV background (Table (Table1)1) by independent deletion of capA or capO. For both mutants, the loss of capsule led to survival levels in human blood that were similar to the level observed for wild-type MRSA (Figure (Figure44).
Alterations in the staphylococcal cell surface have been linked to antimicrobial peptide resistance, predominantly related to changes in cell surface charge . In addition, previous studies have demonstrated a link between staphylococcal resistance to human innate host-defense peptides, such as thrombin-induced platelet microbicidal proteins (tPMPs), and persistent S. aureus infection . Resistance to 2 human antimicrobial peptides, hNP-1 and hBD-2, was therefore analyzed in the clinical and mutant strains. The clinical SCV JKD6229 was more resistant than the parent wild-type strain JKD6210 to both peptides (Figure (Figure5).5). The rpoB and relA mutations each contributed to reduced susceptibility to both antimicrobial peptides, and both mutations in combination (strain BPH1047) demonstrated a more resistant phenotype than either mutant alone for hNP-1. Repair of the rpoB mutation essentially reversed the phenotype. In addition, both acapsular strains demonstrated enhanced susceptibility to peptide as compared to the SCV strain JKD6229, particularly for hNP-1.
To validate the results from radial diffusion assay and investigate the mechanisms of peptide killing, flow cytometric analysis was used, after staining of bacterial cells with proprium iodide, to determine the impact of peptides on membrane function. This demonstrated a significant reduction in permeabilization for the clinical SCV and rpoB or relA mutant–containing strains after hBD-2 treatment, with reversal of this effect with rpoB repair or capsule deletion (Figure (Figure6).6). However, for hNP-1, despite a significant increase in resistance in the relA mutant strain, no reduction in cell permeabilization was demonstrated, indicating another mechanism of hNP-1 resistance in this mutant.
In the present study, we have demonstrated a significant dual phenotype for the clinical SCV strain JKD6229, in which attenuated virulence was associated with in vivo persistence (Figure (Figure1).1). The RpoB H481Y and the RelA F128Y mutations contained in the SCV strain contributed to reduced virulence (Figure (Figure2)2) through transcriptional changes (Figure (Figure3),3), including marked upregulation of capsule biosynthesis, that could be linked with resistance to human innate immune responses (Figures (Figures444–6). Two independent acapsular mutants in the SCV strain (JKD6229) also demonstrated that enhanced capsule expression, which is linked with point mutations in rpoB and relA, was a major contributor to the observed phenotypes.
The H481Y mutation in RpoB, the β-subunit of RNA polymerase, is a common mutation leading to high-level rifampicin resistance in various bacterial species, including S. aureus [17, 18, 31–33]. The fitness costs and compensatory responses resulting from rpoB point mutations in S. aureus is an area of significant interest [18, 34, 35]. Several studies have generally described reduced in vitro fitness of rifampicin-resistant mutants when using culture competition assays . However, a recent study has demonstrated enhanced in vivo survival of rifampicin-resistant mutants, compared with a wild-type strain, in a biofilm model . Here, we have provided a biological explanation for these findings by demonstrating that, although a virulence fitness cost is incurred in S. aureus harboring the RpoB H481Y mutation, other compensatory changes in the organism lead to features promoting persistent in vivo survival. Recent findings show that rpoB mutations, including the H481Y mutation, also promote reduced vancomycin susceptibility in S. aureus . Thus, there is a dual survival advantage arising from this single mutation, through both increased resistance to antimicrobials and innate immune responses. It is likely no coincidence that unusually persistent S. aureus infections are associated with rifampicin resistance [11–13, 15].
Another mutation present in the clinical SCV strain, RelA F128Y, also led to similar phenotypic changes promoting in vivo persistence, particularly in vitro resistance to human defensins (Figure (Figure55 and and6).6). Both the rpoB and the relA mutations led to significant global transcriptional changes as compared to wild-type (Figure (Figure3),3), including changes in genes encoding mediators of central metabolism, global regulation, and virulence-associated genes. In this study, we also constructed an rpoB/relA double mutant to determine whether this genotype fully restored the growth and/or virulence phenotype of the clinical SCV strain. While there was some overlap in the differentially regulated genes in the rpoB and relA mutant, there was an additive effect, with the double mutant producing a transcriptional profile close to that of the clinical SCV (Figure (Figure3).3). However, the virulence of the double mutant was not reduced to the levels of the clinical SCV, indicating that other factors are responsible for the full SCV phenotype.
Here, we have also gained insight into the mechanisms of innate immune resistance and enhanced in vivo persistence by generating acapsular mutants in the SCV strain JKD6229. Capsule deletion reversed the blood killing and antimicrobial peptide resistance of the SCV. The overproduction of both staphylococcal microcapsule types (CP5 and CP8) has been shown to increase virulence in a number of animal models . Enhanced bacterial recovery in blood (up to 4 hours) and deep organs (up to 48 hours) has been demonstrated in a mouse bacteremia model when capsule is overexpressed . Here, we have demonstrated in vivo persistence up to 7 days after intravenous injection in the clinical SCV strain JKD6229. The impact of the rpoB and relA mutations on innate immune responses suggest they promote this phenotype. However, the persistence of these mutants was not tested because mortality remained too high in mice infected with these mutants. Staphylococcal capsule also impairs phagocytosis and reduces phagocytic killing [37, 39, 40]. Our findings support a similar response in the rpoB mutant in the whole-blood killing assay. It is noteworthy that the capsule deletion strains demonstrated whole-blood killing profiles similar to that of the wild-type strain, suggesting that the basal capsule production in this strain was insufficient to impact whole-blood killing in this assay. Notably, upregulation of agr was also demonstrated in the mutant and clinical SCV strain in this study, a transcriptional feature more in keeping with enhanced virulence due to promotion of exotoxin secretion. The β-type PSMs, which are controlled by agr , were also upregulated in the array analysis (Figure (Figure3).3). While the α-type PSMs were not represented on the array slide, we would also expect enhanced expression of these PSM types because of the importance of agr in the control of their expression . Recent data demonstrating the effect of enhanced PSM expression on phagocytosis  suggest that the enhanced PSM expression in the rpoB and relA mutants might contribute to intracellular survival.
The important role played by antimicrobial peptides in the innate immune response is increasingly recognized . Along with the clinical SCV strain, both the rpoB and relA mutants demonstrated enhanced resistance to antimicrobial peptides (Figure (Figure5).5). Other studies have demonstrated a link between antimicrobial peptide resistance and persistent infection isolates, including resistance to hNP-1  and tPMP , but mechanisms have not been defined. Antimicrobial peptide bactericidal mechanisms are complex and involve multiple factors, such as membrane permeabilization and changes in the composition of the cell wall . Resistance to antimicrobial peptides in S. aureus has been found to correlate with an increase in surface positive charge, predominantly mediated by dltABCD-regulated d-alanine modification of cell wall teichoic acids , or an mprF-mediated increase of l-lysine modification of phosphatidylglycerol . Deletion of capsule genes in the clinical SCV strain JKD6229 resulted in increased susceptibility to hNP-1 and restored cell permeabilization to wild-type levels (Figure (Figure6).6). A role for S. aureus capsule in AMP resistance has not been previously reported. However, it has been demonstrated that the purified capsule of gram-negative bacilli, including Klebsiella species  and the gram-positive pathogen Streptococcus pneumoniae, can limit the interaction of the AMPs with the bacterial surface [47, 48]. Likewise, an electrostatically charged polysaccharide capsule of the staphylococcus cell envelope may act as a mechanical barrier, charge sink, or decoy mechanism of peptide resistance. Thus, on the basis of current findings, we propose that enhanced capsule expression is a novel mechanism promoting AMP resistance and persistent S. aureus infection.
In summary, we have described a link between S. aureus antimicrobial resistance, virulence, and host immune responses. These effects are at least partly driven by excess capsule expression and can be induced by single-nucleotide substitutions in rpoB and relA. This research highlights multiple adverse effects of inappropriate antimicrobial therapy, leading not only to antimicrobial resistance but to effects on host-pathogen interactions and the potential to promote persistent infection.
Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.
Acknowledgments.We thank The Institute for Genomic Research, for supplying microarray slides; and Jessica Porter, for expert technical assistance.
Financial support.This work was supported by the National Health and Medical Research Council, Australia, and by the National Institutes of Health (grant 5R01-AI39001 to M. R. Y.).
Potential conflicts of interest.M. R. Y. is a founder and owns equity in NovaDigm Therapeutics, which is developing vaccine technologies against Staphylococcus and other human pathogens. A. Y. P. has been to 1 advisory board meeting for Abbott Molecular and Ortho-McNeil-Janssen and has received a speaker's honorarium from AstraZeneca and Merck, Sharp, and Dohme for 1 presentation each. All other authors report no potential conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed